A Phoneme Harmonic Generator
In the paper we present a new speech harmonic generator that retains the mutual harmonic amplitude ratio. This ratio is important for retaining the speaker timbre in the synthesized signal. The fourth order quasipolynomial model is chosen as the harmonic model in time domain. A new amplitude and phase determining algorithm is derived and its stepwise form presented. A harmonic generator is excited by dynamic pseudo-periodic input sequences. An example of the synthesized harmonic of the vowel “a” and its comparison with the true sound signal harmonic is given.